Research On Product Space Network Dynamic Community Detection And Evolution Analysis

The type of product that a country exports has great impacts on subsequent economic performance. Combined with complex network, the theory of product space uses a network to represent capabilities of products and factor similarities between products, and exhibits the connection of products and the evolution of network structure in a visual way, which provide a new perspective for the research of countries' export structure transformation. Thus, analyzing product space with complex network and community detection methods is helpful for finding out key nodes, exploring characteristics of product space network and tracking the development trend of product and industry. These information can make new suggestion for export structure transformation, industry upgrading and keeping high speed of economic development. This paper aims to research on dynamic product space network community structure discovery and dynamic community evolution detection, then the result from which is applied to analyze and evaluate the export structure of China. The main work is as follows:(1) This paper puts forward an algorithm of community detection for dynamic product space network based on its structural features. Considering the historic community structure, the algorithm combines related international trade theory and complex network community discovery methods. By the process of choosing representative nodes, forming initial network community, optimizing community structure and merging communities, this method can find the community structure which is temporal smoothness. Applying the algorithm to detect dynamic community structure of 15 product space networks from 1998 to 2012, it is proved that this method can find partition result in different periods according to product similarity of factor endowment.(2) This paper proposes a method for analyzing dynamic product space network community evolution based on each community's representative node. This method uses community representatives and calculates match function values of two communities between adjacent time stamps to find each community's predecessor and successor. Then define community evolutionary events which include surviving (grown and shrunken), born, vanished, merged and spilt based on the relationship between communities from different time stamps and the change of community size. Finally, summarize the community evolutionary paths. This method is used to analyze the product space community evolution from 1998 to 2012, especially the evolutionary paths and development trend of electronics, clothes and chemicals.(3) With the information from the result of dynamic product space community detection and community evolution, this paper analyzes China's export structure from 1998 to 2012 respectively in the whole, individuals and communities aspect. The result shows that China's export structure has upgraded during 1998 to 2012 and chemicals should be treated as key development orientation. The conclusion provides practical suggestions for subsequent economic performance of China.